Re: [PATCH 0/5] vt: get rid of worst cases of __put_user/__get_user
On Mon, Jun 05, 2017 at 07:13:50AM +0100, Al Viro wrote: > On Sun, Jun 04, 2017 at 12:42:52AM +0900, Greg Kroah-Hartman wrote: > > On Sat, Jun 03, 2017 at 09:32:55AM +0200, Adam Borowski wrote: > > > Hi! > > > In a recent discussion, Linus and Al Viro said quite a bit of expletives > > > about __put_user() and __get_user(), that it's a bad interface that's > > > almost always the wrong thing to use: > > > https://marc.info/?l=linux-kernel&m=149463725626316&w=2 > > > https://marc.info/?l=linux-kernel&m=149465866929092&w=2 > > > > > > Here's a few patches applying the lessons from that discussion to vt. > > > None of the uses is performance-critical, but at least we get a nice bit > > > of code simplification. And, it's a start of manual review + conversion > > > that Al Viro wants. > > > > Ah, nice work, at first glance these all look good to me. I'll queue > > them up on Monday. > > Could you put that into a separate no-rebase branch? Or I could do that > in vfs.git, for that matter... Yes, here's a tag/branch for you to pull from that will not go away until 4.13-rc1. The following changes since commit 3c2993b8c6143d8a5793746a54eba8f86f95240f: Linux 4.12-rc4 (2017-06-04 16:47:43 -0700) are available in the git repository at: git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty.git/ tags/vt_copy_cleanup_tag for you to fetch changes up to f8564c93e0907651e21d586920e629227bb0d024: vt: drop access_ok() calls in unimap ioctls (2017-06-09 11:07:36 +0200) vt: copy/from_to cleanup for vt code for Al to pull from. Signed-off-by: Greg Kroah-Hartman Adam Borowski (5): vt: use copy_from/to_user instead of __get/put_user for scrnmap ioctls vt: fix unchecked __put_user() in tioclinux ioctls vt: use copy_to_user instead of __put_user in GIO_UNIMAP ioctl vt: use memdup_user in PIO_UNIMAP ioctl vt: drop access_ok() calls in unimap ioctls drivers/tty/vt/consolemap.c | 56 + drivers/tty/vt/vt.c | 6 ++--- drivers/tty/vt/vt_ioctl.c | 8 --- 3 files changed, 19 insertions(+), 51 deletions(-)
Re: [PATCH 0/5] vt: get rid of worst cases of __put_user/__get_user
On Mon, Jun 05, 2017 at 07:13:50AM +0100, Al Viro wrote: > On Sun, Jun 04, 2017 at 12:42:52AM +0900, Greg Kroah-Hartman wrote: > > On Sat, Jun 03, 2017 at 09:32:55AM +0200, Adam Borowski wrote: > > > Hi! > > > In a recent discussion, Linus and Al Viro said quite a bit of expletives > > > about __put_user() and __get_user(), that it's a bad interface that's > > > almost always the wrong thing to use: > > > https://marc.info/?l=linux-kernel&m=149463725626316&w=2 > > > https://marc.info/?l=linux-kernel&m=149465866929092&w=2 > > > > > > Here's a few patches applying the lessons from that discussion to vt. > > > None of the uses is performance-critical, but at least we get a nice bit > > > of code simplification. And, it's a start of manual review + conversion > > > that Al Viro wants. > > > > Ah, nice work, at first glance these all look good to me. I'll queue > > them up on Monday. > > Could you put that into a separate no-rebase branch? Or I could do that > in vfs.git, for that matter... If you just want to take these in vfs.git feel free to, I don't mind at all: Acked-by: Greg Kroah-Hartman that's probably the easiest. thanks, greg k-h
Re: [PATCH 0/5] vt: get rid of worst cases of __put_user/__get_user
On Sun, Jun 04, 2017 at 12:42:52AM +0900, Greg Kroah-Hartman wrote: > On Sat, Jun 03, 2017 at 09:32:55AM +0200, Adam Borowski wrote: > > Hi! > > In a recent discussion, Linus and Al Viro said quite a bit of expletives > > about __put_user() and __get_user(), that it's a bad interface that's > > almost always the wrong thing to use: > > https://marc.info/?l=linux-kernel&m=149463725626316&w=2 > > https://marc.info/?l=linux-kernel&m=149465866929092&w=2 > > > > Here's a few patches applying the lessons from that discussion to vt. > > None of the uses is performance-critical, but at least we get a nice bit > > of code simplification. And, it's a start of manual review + conversion > > that Al Viro wants. > > Ah, nice work, at first glance these all look good to me. I'll queue > them up on Monday. Could you put that into a separate no-rebase branch? Or I could do that in vfs.git, for that matter...
Re: [PATCH 0/5] vt: get rid of worst cases of __put_user/__get_user
On Sat, Jun 03, 2017 at 09:32:55AM +0200, Adam Borowski wrote: > Hi! > In a recent discussion, Linus and Al Viro said quite a bit of expletives > about __put_user() and __get_user(), that it's a bad interface that's > almost always the wrong thing to use: > https://marc.info/?l=linux-kernel&m=149463725626316&w=2 > https://marc.info/?l=linux-kernel&m=149465866929092&w=2 > > Here's a few patches applying the lessons from that discussion to vt. > None of the uses is performance-critical, but at least we get a nice bit > of code simplification. And, it's a start of manual review + conversion > that Al Viro wants. Ah, nice work, at first glance these all look good to me. I'll queue them up on Monday. greg k-h
